[updated 22.12.05 by Dave]

     How to run revamped analysis code :
     -----------------------------------

      linux> source ~cdfsoft/cdf2.cshrc
      linux> setup cdfsoft2 6.1.2
      linux> newrel -t 6.1.2 deuterons_612
      linux> cd deuterons_612
      linux> addpkg -h Deuteron
      linux> root.exe
      root[] .x Deuteron/analysis_scripts/driver.C

      Currently this runs the analysis module "YieldAnalysis" over min-bias data. 
      The ROOT file containing the output is "results/dummy.root" 


     Ntuples/log-files/lumi-files are on the UK CAF disk :
     -----------------------------------------------------
  
       fcdfdata091.fnal.gov:/cdf/scratch/cdfdata/dwaters/deuteron

• [1] dE/dx for positive tracks (25M event sample) : eps ; gif
• [2] dE/dx for negative tracks (25M event sample) : eps ; gif
• [3] dE/dx for negative tracks (25M event sample, showing mass binning) : eps ; gif
• [4] mass distribution for positive tracks : eps ; gif
• [5] mass distribution for negative tracks : eps ; gif
• [6] d0 for D (black) overlayed on d0 for protons (red) : eps ; gif
• [7] d0 for anti-D (black) overlayed on d0 for anti-protons (red) : eps ; gif
• [8] TOF vs log(p) (positive tracks; proton & deuteron curves overlayed) : eps ; gif
• [9] TOF vs log(p) (negative tracks) : eps ; gif
• [10] TOF ratio (TOF compared to that for a proton) vs log(p) (positive tracks) KEY : deuteron curve (red); kaon curve (green); pion curve(blue) : eps ; gif
• [11] TOF ratio (TOF compared to that for a proton) vs log(p) (negative tracks) KEY : deuteron curve (red); kaon curve (green); pion curve(blue) : eps ; gif
• [12] TOF mass distribution for positive tracks : eps ; gif
• [13] TOF mass distribution for negative tracks : eps ; gif

• H1 anti-deuteron paper
• PHENIX deuteron paper
• Deuteron cross-section data

• [ANKUSH ???] Complete the study of Dbar promptness. Plot beam corrected impact parameter for tracks with & without silicon hits. Show that Dbar is consistent with being prompt (may require comparison with pbar, or a better control sample of tracks that we know to be prompt). See if the impact parameter distribution for D is consistent with having two contributions - a prompt and a non-prompt component. If we can count or fit the prompt D component, is it consistent with the number of Dbars ? Plot z0 for D and Dbar and show that Dbar is close(r) to expected distribution.


• [DAVE ???] Create some large samples of single p, pbar, D, Dbar in the momentum range 0 < p/M < 1.5. Eta range (|eta|<1.0) and realsitic z0 distribution. MC Production Instructions.


• [FARRUKH ???] Try to fit the dE/dx mass distributions with a combination of peaks for p, D (pbar, Dbar) plus some background shapes. Should we do a binned fit or unbinned fit ? If binned, bins of log(M) rather than M ? [I think dE/dx resolution is more constant in log(M) than M]. Think about how to do the same for ToF and then do a combined fit.


• [FARRUKH ???] Figure out the luminosities of the samples we have run over : JET-20 including prescale factors; MINBIAS (fixed rate rather than fixed prescale I think). We should have a macro which uses the luminosity log files from the ntupling jobs (these are stored in the same diectories as the ntuples themselves), or perhaps just a run list, together with knowledge of all the prescale factors etc. so that we can accurately compute integrated luminosities.


• [DAVE ???] Further investigate the ToF : how to generate the mass distribution similar to that for dE/dx - need to be able to compare measured and predicted flight times in a more sensible way. Establish which range of (p/M) ToF will do better than, or at least overlap with, dE/dx.


• [ANKUSH ???] For the dE/dx part of the analysis, split the data into two periods - before the COT problems and after the COT was fixed. Do the dE/dx distributions look similar before & after or do we need to find out if different "universal curves" apply to the two periods ? Come up with a definitive run-list for which we can trust dE/dx and put this into the analysis macro so we know we're only making this measurement using reliable data. May require speaking to Stefano Giagu and the dE/dx group.


• [ANKUSH ???] Strip off a few events containing D and Dbar and let's look at them in the event display - it's a very good way to spot reconstruction pathologies, or the presence of unexpected backgrounds.


• [LONGER TERM] Optimise the track cuts. For example, make a tighter cut on the number of COT hits to improve dE/dx resolution and hence p-D separation, but at the expense of track statistics.


• [LONGER TERM] Figure out our pbar and Dbar track finding efficiencies. Effects to consider are (i) fall off of tracking efficiency at small speed (beta*gamma ~ 0.4 ?). (ii) lost Dbar tracks due to material interactions. (iii) other sources of inefficiency ?


• [LONGER TERM] Figure out exactly how we extract the coalescence parameters and compare with HERA/Herwig.